Interobserver variability in the application of a proposed histologic subclassification of endocervical adenocarcinoma

Abstract

A histologic pattern-based system of risk stratification for endocervical adenocarcinoma has been recently proposed on the basis of tumor-stroma interface and lymph-vascular invasion. The key utility of the system lies in separating cases with very low risk for nodal metastases (pattern A) from those with higher risk (patterns B and C), which may alter the treatment approach. In this study, we determine the reproducibility of applying this system among gynecologic pathologists from 2 institutions using blinded review of 49 adenocarcinomas from 2003 to 2013. κ values and pairwise differences are calculated for the proposed 3-tier system (patterns A, B, and C) as well as a modified version comparing pattern A versus patterns B and C combined (2-tier system). Consensus diagnosis for the 3-tier system is reached in 50% of cases, with majority of κ values indicating fair to almost perfect agreement (range, 0.24 to 0.84). When condensed to 2 tiers, consensus is reached in 81.3% of cases with κ values showing modest improvement (range, 0.33 to 0.92). Pairwise difference analysis reveals diagnosis trends for specific pathologists on the 3-tier system that decrease with 2 tiers. Interpretive variability may be of practical significance in application of the proposed 3-tier pattern-based approach to endocervical adenocarcinoma. Additional studies with larger patient cohorts are needed to confirm the negligible risk for lymph node involvement seen in pattern A patients and to further evaluate the applicability of this new classification system.